With the widely used 50% threshold, sensitivity is high, but specificity is low in detecting viable myocardium on 201Tl SPECT. In this study, we sought to identify the best threshold for semiquantitative 201Tl analysis.
Methods: Rest-redistribution 201Tl SPECT was performed in 46 patients with chronic coronary artery disease before and after myocardial revascularization. Regional function was evaluated by two-dimensional echocardiography before and after myocardial revascularization using a 3-point scale (1 = normal, 2 = hypokinetic, 3 = a/dyskinetic). Myocardial segments with abnormal systolic function were defined as viable if the systolic function score decreased > or = 1 after myocardial revascularization. A second group of 12 patients with chronic coronary artery disease constituted the validation population. Sensitivity-specificity curves, as well as receiver operating characteristic curves, for rest and redistribution mages were generated by varying the 201Tl uptake threshold.
Results: A 65% threshold uptake using resting images was found to be the best for detecting a/dyskinetic segments that improve after myocardial revascularization from those that do not improve. Sensitivity was lower with a 65% threshold (75%) than with a 50% threshold (90%, p < 0.05), but specificity was higher (76% versus 26%, p < 0.05) resulting in better accuracy (76% versus 57%, p < 0.05) and positive predictive value (77% versus 55%), while the negative predictive value was not different (69% versus 75%, p not significant). The area under the receiver operating characteristic curve was significantly (p < 005) larger for rest (0.80 +/- 0.05) as opposed to redistribution (0.72 +/- 0.05) images. Similar results were obtained in a subgroup of patients with low ejection fraction. Significant correlations between the percentage of revascularized viable segments and both the change in ejection fraction and in postrevascularization ejection fraction were found. When these findings were applied in the validation group, a gain in specificity, accuracy and positive predictive value was obtained with the 65% threshold compared with the 50% threshold.
Conclusion: This study demonstrated that analysis of resting images and use of the 65% 201Tl uptake threshold is preferable for separating viable from not viable dyssynergic myocardial segments in patients with chronic coronary artery disease.